Matthieu Boffety

Contrast optimization in broadband passive polarimetric imaging.

Polarimetric imaging is often performed using light with a narrow spectrum for the sake of polarization measurement accuracy. However, due to the use of narrowband filters, this reduces the amount of light entering the system and thus the signal-to-noise ratio. This may not be the best choice for target detection applications, where a high target contrast is required rather than polarimetric accuracy. We address contrast optimization for broadband passive polarimetric imaging. We show through simulation and experiments that polarimetric contrast can be significantly increased by broadening the spectrum of analyzed light. In addition, we show that the contrast can be optimized by taking into account the spectral dependence of the scene and of the polarization analysis devices.

Comparison of different active polarimetric imaging modes for target detection in outdoor environment.

We address the detection of manufactured objects in different types of environments with active polarimetric imaging. Using an original, fully adaptive imager, we compare several imaging modes having different numbers of polarimetric degrees of freedom. We demonstrate the efficiency of active polarimetric imaging for decamouflage and hazardous object detection, and underline the characteristics that a polarimetric imager aimed at this type of application should possess. We show that in most encountered scenarios the Mueller matrices are nearly diagonal, and sufficient detection performance can be obtained with simple polarimetric imaging systems having reduced degrees of freedom. Moreover, intensity normalization of images is of paramount importance to better reveal polarimetric contrast.

Analysis of the depth resolution limit of luminescence diffuse optical imaging

We introduce a methodology to determine quantitatively the depth resolution limit in luminescence diffuse optical imaging. The approach is based on a Cramer-Rao statistical analysis, a noise model, and calculations of photon transport in tissues. We illustrate the method in the case of luminescence imaging in a brain-skull model, showing its potential applications in molecular imaging on small animals.

Optimal configuration of static polarization imagers for target detection.

We determine the set of analysis states of a static Stokes imager that maximizes target detection performance for the least favorable target-background polarimetric configuration. By using a minimax approach, we demonstrate that the optimal choice consists of four analysis states forming a regular tetrahedron in the Poincaré sphere. We also show that the value of the contrast in the best of the four Stokes channels is, in the worst case, equal to one third of that provided by a fully adaptive polarimetric imager. Static Stokes imagers thus constitute an attractive solution in applications where limited loss of discrimination ability can be tolerated.

Identification of precancerous lesions by multispectral gastroendoscopy

Gastric cancer is one of the fifth most deadly cancers worldwide. Nowadays the diagnosis is performed through gastroendoscopy under white light and histological analysis. However, the precancerous lesions are multifocalized and present low differences with respect to healthy tissue. Several systems have been proposed based on light tissue interaction to improve the visualization of malignancies. However, these systems are limited to few wavelengths. In this paper, we propose a minimally invasive technique based on multispectral imaging and a methodology to identify malignancies in the stomach. We developed a multispectral gastroendoscopic system compatible with current gastroendoscopes, where only the illumination is changed. The spectra are extracted from the acquired multispectral images in order to compute statistical features that are used to classify the data in two classes: healthy and malignant. The features are ranked by pooled variance t test to train three classifiers. Neural networks using generalized relevance learning vector quantization, support vector machine (SVM) with a Gaussian kernel and k-nn are evaluated using leave one patient out cross-validation. Taking into consideration the data collected in this work, the quantitative results from the classification using SVM show high accuracy and sensitivity using a low number of features. These results show the ability to discriminate malignancies of the gastric tissue. Therefore, multispectral imaging could help in the identification of malignancies during gastroendoscopy.

Performance comparison of fully adaptive and static passive polarimetric imagers in the presence of intensity and polarization contrast.

We address the comparison of contrast improvement obtained with a fully adaptive polarimetric imager and the best channel of a static polarimetric imager in the presence of both intensity and polarization differences between the target and the background. We develop an in-depth quantitative study of the performance loss incurred by a static imager compared to a fully adaptive one in this case. These results are useful to make a well-informed choice between these two polarimetric imaging architectures in a given application.

Fundamental limits of target detection performance in passive polarization imaging

We quantitatively determine the target detection performance of different passive polarization imaging architectures perturbed by signal-independent detection noise or signal-dependent Poisson shot noise. We compare the fully adaptive polarimetric imager and the best channel of a static polarimetric imager, and in each case, we compare the use of a polarizer and a polarizing beam splitter as the polarization analyzing device. For all these configurations, we derive a closed-form expression of the target/background separability and quantify the performance gain brought by polarization imaging compared to standard intensity imaging. We show in particular that all the considered polarimetric imaging configurations but one require a minimum value of the polarimetric contrast in order to outperform intensity imaging. The only configuration that always performs better than intensity imaging uses a polarizing beam splitter in the presence of background shot noise. These results are useful in evaluating the fundamental limits of the gain brought by polarization imaging and determining, in practice, which type of imaging architecture is preferable for a given application.

Improving target discrimination ability of active polarization imagers by spectral broadening.

Active polarization imagers using liquid crystal variable retarders (LCVR) usually operate at one given wavelength for the sake of polarimetric accuracy. However, this often requires to use narrowband filters which reduces the amount of light entering the system and thus the signal-to-noise ratio. For applications where good target/background discriminability (contrast) is required rather than polarimetric accuracy, this may not be the best choice. In this Article, we address contrast optimization in the case of broadband active polarimetric imaging for target detection applications. Through numerical and experimental studies, we show that broadening the spectrum of the light entering the system can increase the contrast between two regions of a scene. Furthermore, we show that this contrast can be further increased by taking into account the spectral dependence of the scene and of the polarimetric properties of the imaging system in the optimization of the measurement procedure.

Optimal configuration of static Mueller imagers for target detection

We investigate the target detection performance of static Mueller imagers that implement a fixed number of illumination and analysis polarization states. Using a maximin approach, we demonstrate that the optimal sets of measurement vectors consist in regular tetrahedra on the Poincaré sphere and that, in this case, the obtained target/background contrast has a very simple expression. We then derive a universal lower bound on the contrast ratio between the best channel of a static imager and a fully adaptive one, and in a special case of practical interest, we demonstrate that this ratio is bounded and always larger than 1/9. This is very important in practice since static imagers are much easier to build and operate. Our results show that they constitute a good alternative where ultimate contrast improvement is not necessary.

Infrared active polarimetric imaging system controlled by image segmentation algorithms: application to decamouflage

We describe an active polarimetric imager with laser illumination at 1.5 µm that can generate any illumination and analysis polarization state on the Poincar sphere. Thanks to its full polarization agility and to image analysis of the scene with an ultrafast active-contour based segmentation algorithm, it can perform adaptive polarimetric contrast optimization. We demonstrate the capacity of this imager to detect manufactured objects in different types of environments for such applications as decamouflage and hazardous object detection. We compare two imaging modes having different number of polarimetric degrees of freedom and underline the characteristics that a polarimetric imager aimed at this type of applications should possess.